AI-Enabled Analysis of 3-D CT Scans for Diagnosis of COVID-19 & its Severity
Dimitrios Kollias, Αναστάσιος Αρσένος, Stefanos Kollias
Abstract
This paper describes the 3rd COVID-19 Competition, taking place in the AI-enabled medical image analysis (AIMIA) Workshop of the 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2023). The 3rd COVID-19 Competition is a continuation of the Competitions held at ECCV 2022 and ICCV 2021 Conferences, and aims to tackle the challenges of whole slide image and CT/MRI/X-ray analysis/processing and to identify research opportunities in the context of Digital Pathology and Radiology/COVID19. The 3rd COVID-19 Competition consists of two Challenges targeting COVID19 detection and COVID19 severity detection. Both Challenges are based on an extended version of the database used in the 1st and 2nd COV19D Competitions, the COV19-CT-DB database, which includes chest CT scan series. A large part of the COV19-CT-DB database is annotated for COVID-19 detection and consists of 8,0003-D CT scans. About 1,0003-D CT scans of the database are also annotated with respect to four COVID-19 severity conditions. Both parts have been split in training, validation and test datasets. These are used for training and validation of machine learning models, as well as for evaluation. The paper further describes the baseline methods for the 3rd COVID-19 Competition, which are deep learning approaches, based on CNN-RNN networks. Their performance on detecting the existence and the severity of COVID-19 is reported.